The discussion of power law distributions reminded me of an article I once read on homelessness and how the un-intuitiveness of this kind of distribution in certain situations may lead to woefully ineffectual responses even when much better ones can be derived. The article, written by Malcolm Gladwell for the New Yorker in 2006, and re-published on his website there, delves into the real population and cost structures of homelessness.
The picture of homelessness in most of our minds is a community of the downtrodden who spend their entire lives on the street. Surely, we think, most of those classified as homeless are incapable of work for whatever reason and wander the urban environment scrounging for a place to sleep and the money for some warm food and (generally assumed to be a prime factor in the homelessness) drugs and alcohol. This is the picture of the “normal” homeless, those we imagine comprise the middle of the bell curve in the normal distribution we form in our minds. So we give some change, or volunteer at a soup kitchen, thinking these are the few small steps we can do to fight the intrenched problem of a large population which seems relegated to a life without structure (physical or existential)
Unfortunately we’ve got it backwards. Gladwell profiles a man named Murray, a chronic homeless with substance abuse problems whose injuries and poor health from a life on the street end up costing tax payers millions of dollars not typically allocated for helping the homeless. The thing is, Murray lives on the tails not the top of the curve of homelessness. Existing in what I guess would be termed the short-tail, Murray and the chronic homeless are actually the small minority of those classified as homeless who spend more than a few days on the street and end up costing the tax payers millions in emergency room services, police activity and jailing. The actual majority of homeless people are those unfortunate to lose housing for a day or two, stay in a shelter for the time before finding assistance of some other sort. The money allocated for traditional homeless aid is usually enough to cover them. It is the small number of chronic homeless who end up draining the system.
This is not a fault of the system, which works fine for the large number of temporary homeless which make up most of the distribution. The problem is a different approach is required for the high cost end of this curve, and that is not an inherently intuitive thought. Pilot programs in Denver have found that a much smaller fraction of the millions a chronic homeless could end up actually costing a city could instead be spent permanently housing and providing the medication and consulting services that these individuals often need. Though it is a difficult concept to swallow in a society generally opposed to “handouts” they have found that the person could be supported for about $15,000 a year, about a third of the cost a chronic homeless racks up from other services and a much larger reduction for some of those at the shortest point on the tail.
I think this example showcases the drastic effects a power law distribution can have on the short end and not just at the long tail. The “famous” homeless are a sub-population of the homeless which clearly needs a different, not intuitive, approach. I’m curious what other problems might be the result not of a lack of effort, but of misplaced effort resulting from a misunderstanding of where the costs are actually coming from.